01
The Long Road To ConveRsion:
The digiTaL PuRChase FunneL
By Andrew Martin, Microsoft’s Atlas Institute
itrct
Most marketers are amiliar with the concept o the “purchase unnel,” – and manyonline marketers plan and execute their campaigns with it in mind. They buy somesites or their ability to generate awareness and reach new customers, and others ortheir ability to persuade consumers to take action and make a purchase. For example,the messaging needed to draw a customer to your site rom a search engine is vastlydierent rom the messaging intended to build top-o-mind awareness on a portal.So knowing when and where consumers are exposed to advertising is invaluable toany media plan attempting to leverage each consumers place in the unnel.Today, the reality is that media channels overlap and multiple sites reach consumersprior to conversion
1
. Yet, the advertising metrics and models developed or onlinemarketers largely ignore the question o when media exposure occurs. By measuringthe last ad seen or clicked by a converter, they ocus entirely on a brie time span atthe bottom o the unnel. Evaluating only a single point o advertising contact over-simplies the delivery and perormance o any media channel.
This Digital Marketing Insight aims to answer the question of exactly when andwhere purchasers see online advertising, and what implications those discoveriesmay have on the process of planning and optimizing media.
1
Strong, Esco. Combined Impact o Search and Display, 2006.
Microsoft’s Atlas Institute
02
Mtly & dt
The data or this DMI comes rom over seventeen million conversions that occurredbetween May and September o 2008. These conversions represent the results o advertising campaigns run by 250 advertisers across thousands o sites. Each advertisertracked both search and display campaigns simultaneously, and only conversions romrevenue-producing (such as sales, lead, or registration)action tags were used. The study considered a ull 90 dayso advertising history prior to each conversion.
M Cmpt or Tm
Most advertisers made their rst contact with convertersmonths prior to the conversion. Figure 1 shows the mediannumber o media touch points at dierent times inusers’ conversion histories. Unsurprisingly, the amounto advertising exposure is quite high in the last ew daysbeore a conversion. Consumers amass a median o 5.5ad events in the nal 48 hours beore conversion alone.But this requency amounts to less than a third o the totalmedia events delivered over three months. Hal o the totalengagements with converters occurred between seven andsixty days prior to conversion, and ater a ull 90-day lookback, converters experienced a median o 18.5 events.The large number o ad exposures consumed prior topurchase may come as a surprise to marketers who areused to discussions o requency that revolve around siteor campaign metrics. Measuring only the last ad in aconversion history conceals the true length o the rela-tionship an advertiser has with each consumer. When we ocus our view on individualconverters’ histories and apply the unnel concept to their ad consumption, we discoverthat their histories are much longer and richer than typically assumed. These resultsconrm other research showing that advertising reaches consumers rom multipleadvertising campaigns and across channels
2
.
T Prc Fl
With such a large number o ads being delivered beore conversion, it’s common thatmost converters are reached by an advertiser across multiple sites. Figure 2 lays out themedian time that dierent categories o publishers reach consumers prior to conver-sion. Notably, aliate networks and search engines oten reach customers or the rsttime just a ew days or hours prior to conversion. But categories like Entertainmentand News tend to reach in-market consumers weeks or months beore conversion, anddeliver their requency over a longer period o time. The reason why these sites reachconsumers at such dierent times becomes clearer when we consider where they might
9060302114721
D a y s B e f o r e C o n v e r s i o n
Median Total Ad Events byTotal Days of History
0 5 10
Median Touchpoints
15 20
2.25.58.410.712.313.816.718.5
Figure 1:Over half of advertisingexposure occurredmore than a weekprior to conversion.
2
Fowler, Jed. The Impact o Overlap, 2006.
03logically reach a consumer who is traveling throughthe sales unnel.The interplay between three actors appears toinfuence when a publisher is likely to reach aconverter: site unction, audience, and advertisermix. Search engines are an interesting example o this phenomena. The inormation gathering andnavigational unctions o search draw an audienceo consumers with a high probability o makinga purchase
3
. Consequently, search has attracteda number o direct response and lead generatingadvertisers pushing messages designed to getconsumers to “act now.”Social media, on the other hand, oten reachesconsumers much earlier. Users o social media visitthose sites in order to connect and share experienceswith riends and amily over the web. Most o theunctionality provided by these sites does not leaddirectly to a commercial transaction. Rather,advertisers nd these publishers valuable or theirreach to a broader audience and ability to buildawareness. As a result, social media sites tend tolive high in their advertisers’ purchase unnels.
implct Fr Cmp Mrmt
When a site reaches a user prior to conversion, it has serious implications or how wemeasure and plan online campaigns. This is especially true when it comes to measuringcampaign ROI using the last-ad standard. Over hal o the views or clicks that arecredited by the last-ad methodology occur within ve days or less rom the time o conversion, even though only a raction o the total ad consumption by convertersoccurs during this period. But more importantly, as Figure 2 shows, only a ew sitecategories dominate the lower reaches o the purchase unnel. Assigning credit to thelast click or view disproportionally avors those ew publishers at the expense o thepublishing majority, whose contributions are spread out over more time.Other common industry practices can urther obscure the contributions a publishermade to a sale. Advertisers whose action tags use shorter view or click windows arealso devaluing many publishers, since publishers that appear in the nal week prior toconversion are oten dierent rom those appearing three or our weeks prior. Thesesites are bought or dierent reasons than the publishers urther down the unnel, suchas to generate awareness or reach a particular audience. Publishers higher in the unnelare valuable sources o new prospects that are used to eed trac to lower-unnel
Photo
M e d i a n D a y s B e f o r e C o n v e r s i o n s
3528211470
City GuideAutoAffiliateGamesKidsAd NetworkPortalNewsLifestyleSocialSportsTechnologyEntertainmentFood/WineMusicReal EstateSearch EnginesTelecomFinanceTravelJobsISPShopping VideoWeatherEmailDirectory
Median Ad Delivery Time by Site Category
3
Brooks, Nico. Paying or Navigational Search, 2007.
Figure 2:Most publishers livein the middle andupper funnel.
Search History:
Result 00 of 00
00 results for result for
p.
Notes
Load more
